基于机器视觉的铜箔表面缺陷检测方法研究

Research on Copper Foil Appearance Defect Detection Method Based on Machine Vision

针对当前锂电池用电解铜箔(锂电铜箔)表面缺陷问题,提出了一种基于改进Canny边缘检测算法的铜箔表面缺陷检测方法。首先,使用中值滤波替代传统Canny算法中的高斯滤波,解决高斯滤波会模糊缺陷边缘的问题;其次,使用迭代法自动获取双阈值,增强阈值自适应性,解决固定阈值适应性差的问题。利用工业相机采集到铜箔图像并上传到上位机中,先对图像进行中值滤波和灰度值缩放变换处理,在去除噪声和增强缺陷信息的同时保留缺陷边缘,然后使用动态阈值分割图像获取感兴趣区域(缺陷ROI),最后用改进Canny边缘检测和形态学闭运算结合的方法提取缺陷边缘,完成缺陷检测。实验结果表明传统Canny算法检测成功率较低,而该方法能够精准提取缺陷边缘信息,检测成功率达到97.08%,实现了对铜箔划痕、裂纹、污点类等表面缺陷的准确检测。

Aiming at the current appearance defects of electrolytic copper foil(lithium copper foil)for lithium batteries,a copper foil appearance defect detection method was proposed based on an improved Canny edge detection algorithm.Firstly,the median filter was used to replace the Gaussian filter in the traditional Canny algorithm to solve the problem that the Gaussian filter would blur the edge of the defect.Secondly,the iterative method was used to automatically obtain the dual threshold value,enhance the threshold value adaptability,and solve the problem of poor adaptability of the fixed threshold value.The copper foil image was collected by an industrial camera and uploaded to the host computer.The image was first subjected to median filtering and gray scale value scaling transformation processing to remove noise and enhance defect information while retaining the edge of the defect,and then dynamic threshold segmentation was used to obtain the image.The region of interest(defect ROI)was finally used to extract the edge of the defect by the combination of improved Canny edge detection and morphological closed operation to complete the defect detection.Experimental results show that the traditional Canny algorithm has a low detection success rate,while the method designed can accurately extract defect edge information,with a detection success rate of 97.08%,and achieve accurate detection of copper foil scratches,cracks,stains and other appearance defects.